According to a June 10 statement, the regulator asked Meta to postpone large language model (LLM) training involving public content adults share on Facebook and Instagram.
The Irish DPC acts as Meta’s lead privacy regulator in the EU and issued the request on behalf of the European Data Protection Authorities (DPAs).
The lack of local information means users would receive a “second-rate experience.”
Meta said:
“This means we aren’t able to launch Meta AI in Europe at the moment.”
Meta said it was “disappointed” by the request and called it a “step backward” for AI competition and innovation in the EU. The firm added that it is confident it complies with European regulations.
The Irish DPC endorsed the training pause. It said it “welcomes the decision” and will continue working with Meta following earlier “intensive engagement.”
European privacy non-profit NOYB said that complaints from itself, other organizations such as the Norwegian Consumer Council, and certain DPAs led the Irish DPC to issue the request to pause Meta’s activities.
Situation developed over months
Meta said on June 14 that it had informed European DPAs of its activities since March.
However, the situation developed after Meta started informing users of its AI training activities through more than two billion in-app notifications and emails on May 22.
NOYB sent complaints to 11 DPCs on June 6.
On June 10, Meta acknowledged consultations with the Irish Data Protection Commission (DPC) and said it had integrated feedback to ensure its AI training complied with AI laws.
Meta noted that Google and OpenAI already use AI for user data training. The firm expects to comply with EU General Data Protection Regulation (GDPR) laws on the legal basis of “Legitimate Interest,” a flexible provision that others in the AI industry have relied on.